In a wafer fabrication facility, semiconductor processes are typically performed on groups of 25 to 50 wafers. Each such group of wafers is termed a "lot". Semiconductor processing is thus generally tracked, controlled and analyzed at the lot level and wafers in a lot are not individually distinguished. Therefore, it is difficult to determine the exact process conditions of each wafer due to process variations within each run. However for the purposes of quality assurance and quality control, it is highly desirable to track individual wafers through the entire semiconductor process. Process validation of each individual wafer becomes possible with wafer tracking to prevent misprocessing. Wafer sequential and positional data also give more meaning to the yield and parametric analyses of equipment and process. Additionally, when wafers can be separately identified, wafer mishandling that results in mixed lots can be easily identified and sorted.
Previously, optical character recognition (OCR) has been used to read alphanumeric markings inscribed on wafers for wafer identification. However, the use of optical character recognition has been expensive and problematic. This method uses video imaging to recognize the alphanumeric characters, thus requiring sophisticated computer hardware and software. Because the alphanumeric string is inscribed on a highly reflective surface, precise lighting and camera angles are crucial to successful reading. In addition, there must not be any movement or vibration of the camera or wafer when character recognition is taking place. Any partial obstruction of the alphanumeric markings may also cause the character recognition to fail.
International Business Machines (IBM) has developed a bar code format BC412 that has been adopted as a standard by Semiconductor Equipment and Materials International (SEMI). IBM employs BC412 bar code inscription on the backsides of the wafers for wafer identification. However, the bar code technology has yet to be successfully applied in a high volume wafer fabrication production environment where wafer tracking is particularly beneficial.
The task of applying the bar code technology to a production environment is not trivial. Each wafer fabrication step alters the surface characteristics of the wafer as well as any inscription used for wafer identification. The inscription must survive ion implantation, diffusion, photolithography, metalization, plasma etch, wet etch, oxide deposition, and any other fabrication process and remain readable.
Accordingly, it is desirable to identify and track individual wafers in a high volume production environment without many of the disadvantages and constraints of optical character recognition.